parser.php

<?php
/*
    Code taken from WordPress and WordPress Importer Plugin. 
    GPLv2 or later
*/

/**
 * WordPress Error API.
 *
 * Contains the WP_Error class and the is_wp_error() function.
 *
 * @package WordPress
 */

/**
 * WordPress Error class.
 *
 * Container for checking for WordPress errors and error messages. Return
 * WP_Error and use {@link is_wp_error()} to check if this class is returned.
 * Many core WordPress functions pass this class in the event of an error and
 * if not handled properly will result in code errors.
 *
 * @package WordPress
 * @since 2.1.0
 */
class WP_Error {
	/**
	 * Stores the list of errors.
	 *
	 * @since 2.1.0
	 * @var array
	 * @access private
	 */
	var $errors = array();

	/**
	 * Stores the list of data for error codes.
	 *
	 * @since 2.1.0
	 * @var array
	 * @access private
	 */
	var $error_data = array();

	/**
	 * Constructor - Sets up error message.
	 *
	 * If code parameter is empty then nothing will be done. It is possible to
	 * add multiple messages to the same code, but with other methods in the
	 * class.
	 *
	 * All parameters are optional, but if the code parameter is set, then the
	 * data parameter is optional.
	 *
	 * @since 2.1.0
	 *
	 * @param string|int $code Error code
	 * @param string $message Error message
	 * @param mixed $data Optional. Error data.
	 * @return WP_Error
	 */
	function __construct($code = '', $message = '', $data = '') {
		if ( empty($code) )
			return;

		$this->errors[$code][] = $message;

		if ( ! empty($data) )
			$this->error_data[$code] = $data;
	}

	/**
	 * Retrieve all error codes.
	 *
	 * @since 2.1.0
	 * @access public
	 *
	 * @return array List of error codes, if available.
	 */
	function get_error_codes() {
		if ( empty($this->errors) )
			return array();

		return array_keys($this->errors);
	}

	/**
	 * Retrieve first error code available.
	 *
	 * @since 2.1.0
	 * @access public
	 *
	 * @return string|int Empty string, if no error codes.
	 */
	function get_error_code() {
		$codes = $this->get_error_codes();

		if ( empty($codes) )
			return '';

		return $codes[0];
	}

	/**
	 * Retrieve all error messages or error messages matching code.
	 *
	 * @since 2.1.0
	 *
	 * @param string|int $code Optional. Retrieve messages matching code, if exists.
	 * @return array Error strings on success, or empty array on failure (if using code parameter).
	 */
	function get_error_messages($code = '') {
		// Return all messages if no code specified.
		if ( empty($code) ) {
			$all_messages = array();
			foreach ( (array) $this->errors as $code => $messages )
				$all_messages = array_merge($all_messages, $messages);

			return $all_messages;
		}

		if ( isset($this->errors[$code]) )
			return $this->errors[$code];
		else
			return array();
	}

	/**
	 * Get single error message.
	 *
	 * This will get the first message available for the code. If no code is
	 * given then the first code available will be used.
	 *
	 * @since 2.1.0
	 *
	 * @param string|int $code Optional. Error code to retrieve message.
	 * @return string
	 */
	function get_error_message($code = '') {
		if ( empty($code) )
			$code = $this->get_error_code();
		$messages = $this->get_error_messages($code);
		if ( empty($messages) )
			return '';
		return $messages[0];
	}

	/**
	 * Retrieve error data for error code.
	 *
	 * @since 2.1.0
	 *
	 * @param string|int $code Optional. Error code.
	 * @return mixed Null, if no errors.
	 */
	function get_error_data($code = '') {
		if ( empty($code) )
			$code = $this->get_error_code();

		if ( isset($this->error_data[$code]) )
			return $this->error_data[$code];
		return null;
	}

	/**
	 * Append more error messages to list of error messages.
	 *
	 * @since 2.1.0
	 * @access public
	 *
	 * @param string|int $code Error code.
	 * @param string $message Error message.
	 * @param mixed $data Optional. Error data.
	 */
	function add($code, $message, $data = '') {
		$this->errors[$code][] = $message;
		if ( ! empty($data) )
			$this->error_data[$code] = $data;
	}

	/**
	 * Add data for error code.
	 *
	 * The error code can only contain one error data.
	 *
	 * @since 2.1.0
	 *
	 * @param mixed $data Error data.
	 * @param string|int $code Error code.
	 */
	function add_data($data, $code = '') {
		if ( empty($code) )
			$code = $this->get_error_code();

		$this->error_data[$code] = $data;
	}
}

/**
 * Check whether variable is a WordPress Error.
 *
 * Looks at the object and if a WP_Error class. Does not check to see if the
 * parent is also WP_Error, so can't inherit WP_Error and still use this
 * function.
 *
 * @since 2.1.0
 *
 * @param mixed $thing Check if unknown variable is WordPress Error object.
 * @return bool True, if WP_Error. False, if not WP_Error.
 */
function is_wp_error($thing) {
	if ( is_object($thing) && is_a($thing, 'WP_Error') )
		return true;
	return false;
}
/**
 * Butchered version of WordPress eXtended RSS file parser implementations
 *
 * @package WordPress
 * @subpackage Importer
 */

/**
 * WordPress Importer class for managing parsing of WXR files.
 */
class WXR_Parser {
	function parse( $file ) {
		// Attempt to use proper XML parsers first
		if ( extension_loaded( 'simplexml' ) ) {
			$parser = new WXR_Parser_SimpleXML;
			$result = $parser->parse( $file );

			// If SimpleXML succeeds or this is an invalid WXR file then return the results
			if ( 'SimpleXML_parse_error' != $result->get_error_code() )
				return $result;
		} 
    }
}

/**
 * WXR Parser that makes use of the SimpleXML PHP extension.
 */
class WXR_Parser_SimpleXML {
	function parse( $file ) {
		$authors = $posts = $categories = $tags = $terms = array();

		$internal_errors = libxml_use_internal_errors(true);
		$xml = simplexml_load_file( $file );
		// halt if loading produces an error
		if ( ! $xml )
			return new WP_Error( 'SimpleXML_parse_error', __( 'There was an error when reading this WXR file', 'wordpress-importer' ), libxml_get_errors() );

		$wxr_version = $xml->xpath('/rss/channel/wp:wxr_version');
		if ( ! $wxr_version )
			return new WP_Error( 'WXR_parse_error', __( 'This does not appear to be a WXR file, missing/invalid WXR version number', 'wordpress-importer' ) );

		$wxr_version = (string) trim( $wxr_version[0] );
		// confirm that we are dealing with the correct file format
		if ( ! preg_match( '/^\d+\.\d+$/', $wxr_version ) )
			return new WP_Error( 'WXR_parse_error', __( 'This does not appear to be a WXR file, missing/invalid WXR version number', 'wordpress-importer' ) );

		$base_url = $xml->xpath('/rss/channel/wp:base_site_url');
		$base_url = (string) trim( $base_url[0] );

		$namespaces = $xml->getDocNamespaces();
		if ( ! isset( $namespaces['wp'] ) )
			$namespaces['wp'] = 'http://wordpress.org/export/1.1/';
		if ( ! isset( $namespaces['excerpt'] ) )
			$namespaces['excerpt'] = 'http://wordpress.org/export/1.1/excerpt/';

		// grab authors
		foreach ( $xml->xpath('/rss/channel/wp:author') as $author_arr ) {
			$a = $author_arr->children( $namespaces['wp'] );
			$login = (string) $a->author_login;
			$authors[$login] = array(
				'author_id' => (int) $a->author_id,
				'author_login' => $login,
				'author_email' => (string) $a->author_email,
				'author_display_name' => (string) $a->author_display_name,
				'author_first_name' => (string) $a->author_first_name,
				'author_last_name' => (string) $a->author_last_name
			);
		}

		// grab cats, tags and terms
		foreach ( $xml->xpath('/rss/channel/wp:category') as $term_arr ) {
			$t = $term_arr->children( $namespaces['wp'] );
			$categories[] = array(
				'term_id' => (int) $t->term_id,
				'category_nicename' => (string) $t->category_nicename,
				'category_parent' => (string) $t->category_parent,
				'cat_name' => (string) $t->cat_name,
				'category_description' => (string) $t->category_description
			);
		}

		foreach ( $xml->xpath('/rss/channel/wp:tag') as $term_arr ) {
			$t = $term_arr->children( $namespaces['wp'] );
			$tags[] = array(
				'term_id' => (int) $t->term_id,
				'tag_slug' => (string) $t->tag_slug,
				'tag_name' => (string) $t->tag_name,
				'tag_description' => (string) $t->tag_description
			);
		}

		foreach ( $xml->xpath('/rss/channel/wp:term') as $term_arr ) {
			$t = $term_arr->children( $namespaces['wp'] );
			$terms[] = array(
				'term_id' => (int) $t->term_id,
				'term_taxonomy' => (string) $t->term_taxonomy,
				'slug' => (string) $t->term_slug,
				'term_parent' => (string) $t->term_parent,
				'term_name' => (string) $t->term_name,
				'term_description' => (string) $t->term_description
			);
		}

		// grab posts
		foreach ( $xml->channel->item as $item ) {
			$post = array(
				'post_title' => (string) $item->title,
				'guid' => (string) $item->guid,
			);

			$dc = $item->children( 'http://purl.org/dc/elements/1.1/' );
			$post['post_author'] = (string) $dc->creator;

			$content = $item->children( 'http://purl.org/rss/1.0/modules/content/' );
			$excerpt = $item->children( $namespaces['excerpt'] );
			$post['post_content'] = (string) $content->encoded;
			$post['post_excerpt'] = (string) $excerpt->encoded;

			$wp = $item->children( $namespaces['wp'] );
			$post['post_id'] = (int) $wp->post_id;
			$post['post_date'] = (string) $wp->post_date;
			$post['post_date_gmt'] = (string) $wp->post_date_gmt;
			$post['comment_status'] = (string) $wp->comment_status;
			$post['ping_status'] = (string) $wp->ping_status;
			$post['post_name'] = (string) $wp->post_name;
			$post['status'] = (string) $wp->status;
			$post['post_parent'] = (int) $wp->post_parent;
			$post['menu_order'] = (int) $wp->menu_order;
			$post['post_type'] = (string) $wp->post_type;
			$post['post_password'] = (string) $wp->post_password;
			$post['is_sticky'] = (int) $wp->is_sticky;

			if ( isset($wp->attachment_url) )
				$post['attachment_url'] = (string) $wp->attachment_url;

			foreach ( $item->category as $c ) {
				$att = $c->attributes();
				if ( isset( $att['nicename'] ) )
					$post['terms'][] = array(
						'name' => (string) $c,
						'slug' => (string) $att['nicename'],
						'domain' => (string) $att['domain']
					);
			}

			foreach ( $wp->postmeta as $meta ) {
				$post['postmeta'][] = array(
					'key' => (string) $meta->meta_key,
					'value' => (string) $meta->meta_value
				);
			}

			foreach ( $wp->comment as $comment ) {
				$meta = array();
				if ( isset( $comment->commentmeta ) ) {
					foreach ( $comment->commentmeta as $m ) {
						$meta[] = array(
							'key' => (string) $m->meta_key,
							'value' => (string) $m->meta_value
						);
					}
				}
			
				$post['comments'][] = array(
					'comment_id' => (int) $comment->comment_id,
					'comment_author' => (string) $comment->comment_author,
					'comment_author_email' => (string) $comment->comment_author_email,
					'comment_author_IP' => (string) $comment->comment_author_IP,
					'comment_author_url' => (string) $comment->comment_author_url,
					'comment_date' => (string) $comment->comment_date,
					'comment_date_gmt' => (string) $comment->comment_date_gmt,
					'comment_content' => (string) $comment->comment_content,
					'comment_approved' => (string) $comment->comment_approved,
					'comment_type' => (string) $comment->comment_type,
					'comment_parent' => (string) $comment->comment_parent,
					'comment_user_id' => (int) $comment->comment_user_id,
					'commentmeta' => $meta,
				);
			}

			$posts[] = $post;
		}

		return array(
			'authors' => $authors,
			'posts' => $posts,
			'categories' => $categories,
			'tags' => $tags,
			'terms' => $terms,
			'base_url' => $base_url,
			'version' => $wxr_version
		);
	}
}